A Best-Estimate Paradigm for Diagnosis of Multiple Failure Transients in Nuclear Power Plants Using Artificial Intelligence

A paradigm for best-estimate diagnosis of reactor transients has been developed. The theoretical approach is based on a modified assumption-based truth maintenance system (ATMS). By incorporating a conflict resolution strategy using expert confidence levels, ATMS is extended. Several software experiments were performed to assess validity of the theory. The software experiment included features for uncertainty management, multiple failure diagnosis in real time, and retrieval of appropriate emergency operating guidelines for accident mitigation. The software experiments demonstrated plausible results for loss of feedwater, loss-of-coolant accident, anticipated transient without scram, and steam generator tube rupture transients.